did you measure elapsed time in gSSP. especially calling the upstream API or DB. so you can nail down the culprit.

7 seconds are too long so most probably network issues

or fetch enormously huge payloads from upstream, while your self hosted version sits the upstream closely so fetched faster

Hi @tafutada777

Im living in Canada and the personal server is hosted in Paris
Regarding the "huge payloads", unfortunately this is not the current issue, the payload is really small, like ~10 fields and only strings (small ones as well)

With insomnia the request tooks like 200ms maximum
On my private server, almost the same, or at least very close
On Vercel takes 5 seconds, but only the first request, once this is one is done, the next ones will be fast, as it should be

thank you for your answer 👌

It sounds like cold start on lambda. Is your getServerSideProps calling an external api or one in your next project?

calling an external API yes

So, usually in production a busy site will keep its functions "warm" except right after redeployment. For testing sites they are often cold starts

Are you having a production issue?

website has been deployed on vercel, with the default configuration, and looks like a production deployment, even if I missed something ?
so yes i'd say it's a production issue

Even if a friend go on the website right after me, he'll get the long first loading as well

Looks like a production deployment in terms of traffic?

Oh no, not yet at least
Website is still "private", so only the devs got the access
Should be the issue ?

I mean, once we'll release it in public and get more traffic, the cold start issue won't be there anymore ? or just after a re-deployment ?

There can also be cold starts with a rapid increase in traffic as more handlers are provisioned, but that is likely to outperform what a self-hosted solution would do

Ok thank you @Asian black bear ! really appreciate, we'll monitor this issue once we'll release the new website!
Thanks again, and have a nice day!

BTW, I have that problem while using nextauth ( since it is using getServersideProps) . In the web application we have like 60 test users and they are using website at least 2 hours a day.

We couldn't improve the page speed. It was on vercel and i also tried to deploy it on AWS Amplify

The website is app.globalturkish.org , its on AWS Amplify, and all of microservices and CDNs are also deployed in frankfurt

If you find any optimization solutions would happy to hear from you !

Another thing to consider is using the edge runtime when possible